AMD Introduces Dynamic Local Mode for Threadripper: up to 47% Performance Gain

[Nelly]

You know what? I think this deserves its own thread, it's news. If you don't like it, don't read it, simple as that.

AMD has made a blog post describing an upcoming feature for their Threadripper processors called "Dynamic Local Mode", which should help a lot with gaming performance on AMD's latest flagship CPUs.



Threadripper uses four dies in a multi-chip package, of which only two have a direct access path to the memory modules. The other two dies have to rely on Infinity Fabric for all their memory accesses, which comes with a significant latency hit. Many compute-heavy applications can run their workloads in the CPU cache, or require only very little memory access; these are not affected. Other applications, especially games, spread their workload over multiple cores, some of which end up with higher memory latency than expected, which results in a suboptimal performance.

The concept of multiple processors having different memory access paths is called NUMA (Non-uniform memory access). While technically it is possible for software to detect the NUMA configuration and attach each thread to the ideal processor core, most applications are not NUMA aware and the adoption rate is very slow, probably due to the low number of systems using such a concept.



In ThreadRipper, using Ryzen Master, users are free to switch between "Local Memory Access" mode or "Distributed Memory Access" mode, with the latter being the default for ThreadRipper, resulting in highest compute application performance. Local Mode on the other hand is better suited to games, but switching between the modes requires a reboot, which is very inconvenient for users.

AMD's new "Dynamic Local Mode" seeks to abolish that requirement by introducing a background process that continually monitors all running applications for their CPU usage and pushes the more busy ones onto the cores that have direct memory access, by adjusting their process affinity mask, which selects which processors the application is allowed to be scheduled on. Applications that require very little CPU are in turn pushed onto the cores with no memory access, because they are not so important for fast execution.



This update will be available starting October 29 in Ryzen Master, and will be automatically enabled unless the user manually chooses to disable it. AMD also plans to open the feature up to even more users by including Dynamic Local Mode as a default package in the AMD Chipset Drivers.

Source: AMD Blog Post / TechPowerup

 

[humbug]

As i understand it, no longer is it necessary to switch between Creator and Gamers mode with Threadripper to give the best performance high threaded applications vs games. "Dynamic Local mode" will do all of this automatically and on the fly without user input.

 

[Panos]

As i understand it, no longer is it necessary to switch between Creator and Gamers mode with Threadripper to give the best performance high threaded applications vs games. "Dynamic Local mode" will do all of this automatically and on the fly without user input.

Yeah. Also raised a question on the official blog if this will work when a user has dual channel not quad channel ram on tr4 system. Hadn't received reply yet.

Given the ram pricing going to be a boon to AMD if it works as it would allow immediately people use their dual channel ram to upgrade to platform.

 

[Hotwired]

So... you get "Game Mode" performance with no fiddling required then?

Or will this be better seeing as it doesn't seem to disable any cores to do it.

 

[B1gbeard]

That seems misleading. Essentially you can already do this switching to game mode? There shouldn't be any performance numbers here, just we have now made it possible to do this on the fly.

 

[blaineoliver]

That seems misleading. Essentially you can already do this switching to game mode? There shouldn't be any performance numbers here, just we have now made it possible to do this on the fly.

Unless I've misunderstood, you don't lose half your CPU cores it need to reboot between applications that benefit from creator mode Vs the ones that benefit from gaming mode.

As for the performance numbers I'm not sure how they stack up compared to the different modes.

 

[B1gbeard]

Unless I've misunderstood, you don't lose half your CPU cores it need to reboot between applications that benefit from creator mode Vs the ones that benefit from gaming mode.

As for the performance numbers I'm not sure how they stack up compared to the different modes.

That seems to be it but at the top of the post AMD suggest it helps with gaming performance and then gives numbers. There is no extra performance here as the CPU already does those numbers in game mode or at least that is what it seems to be lower down the post. This should have been something along the lines of You can now get game mode gaming performance without restarting your PC. Super misleading until you read further down.

 

[Hotwired]

That seems to be it but at the top of the post AMD suggest it helps with gaming performance and then gives numbers. There is no extra performance here as the CPU already does those numbers in game mode or at least that is what it seems to be lower down the post. This should have been something along the lines of You can now get game mode gaming performance without restarting your PC. Super misleading until you read further down.

We need to read the ENTIRE article?

Also you're reading a 3rd party repost with added clickbait. This is the offical post from AMD: https://community.amd.com/community...e-amd-ryzen-threadripper-wx-series-processors

 

[B1gbeard]

We need to read the ENTIRE article?

Still misleading regardless and if another company had done it we'd be talking pitchforks. There can't be a gain when the CPU can already do it. As I said a simple "We now change modes on the fly" would have been better.

 

[Hotwired]

Still misleading regardless and if another company had done it we'd be talking pitchforks. There can't be a gain when the CPU can already do it. As I said a simple "We now change modes on the fly" would have been better.

Check my link. You're not reading an article from AMD here, you're reading TechPowerUp's article using AMD's graphics.

 

[B1gbeard]

We need to read the ENTIRE article?

Also you're reading a 3rd party repost with added clickbait. This is the offical post from AMD: https://community.amd.com/community...e-amd-ryzen-threadripper-wx-series-processors

This quote from the original "In the applications we have tested to date, AMD has observed performance improvements of up to 47% with Dynamic Local Mode enabled. The below diagram shows a variety of games and applications aided by the new feature, and AMD expects other applications that we have not yet analyzed may also benefit."

Further on in that paragraph it says "Even so, it's clear that some processes really take a liking to Dynamic Local Mode and it's quite satisfying to see such a speedup from a new and free feature for your platform" when there is no real performance gain unless these numbers are over and above switching manually.

 

[Hotwired]

I think you're reaching on this seeing as they're showing everything.

For example just above the bit you quoted there was an introduction to what they were about to show...

Thinking back to the 1st Generation AMD Ryzen™ Threadripper processor, AMD Ryzen™ Master was updated to add a toggle for Local Mode or Distributed Mode. These modes tuned the performance of applications that preferred lower memory latency or higher memory bandwidth, respectively. This capability required a system reboot but, according to reviewers like TechSpot, there was a clear performance upside when an application was paired with its most favored mode.

With the “favored modes” in mind, that brings us to today. What if Ryzen™ Threadripper WX Series CPUs could have some sort of “favored mode” to ensure the best performance for both heavy and lightly-threaded apps? What if it could be switched on the fly without a reboot? All of this is possible with a new feature we’re calling Dynamic Local Mode.

Sneaky of them eh.

Also you missed the footnote pointed to exactly where the 47% claim is made which is:

1. Testing by AMD Performance Labs as of 10/4/2018. Results presented in order of Dynamic Local Mode OFF vs. ON (% difference).

 

[B1gbeard]

I think you're reaching on this seeing as they're showing everything.

For example just above the bit you quoted there was an introduction to what they were about to show...



Sneaky of them eh.

Also you missed the footnote pointed to exactly where the 47% claim is made which is:

The paragraphs you have quoted are fine, but there is no performance gain whatever way you look and it and for me anyway the paragraph and chart damages the press release. Remove that and it's fine. They are claiming a performance gain when there simply isn't one irrespective of how they explain how it works. Even in respect to the rest of the article you could still get misled into thinking DLM is doing something extra from that paragraph.

 

[Deleted member 66701]

I don't think it's misleading and I think B1gbeard you're trying to find controversy where there is none.

Its a cool piece of tech from AMD, just shame that they have to do Microsofts job for them.

 

[Panos]

The paragraphs you have quoted are fine, but there is no performance gain whatever way you look and it and for me anyway the paragraph and chart damages the press release. Remove that and it's fine. They are claiming a performance gain when there simply isn't one irrespective of how they explain how it works. Even in respect to the rest of the article you could still get misled into thinking DLM is doing something extra from that paragraph.

Nothing is misleading. Up to now you had to shutdown half your Threadripper to play games. In the case of 2990WX you have to shut down 2/3 of the CPU to do so successfully.
AMD now provides the option to keep your full CPU active, and the AMD API will assess the needs of each process and push it down to cores with no access to Ram if it can.

AMD doesn't claim found some perf improvements magically. But now we don't have to shut down half the CPU to do so

In effect AMD is fixing the Windows Scheduler, Microsoft has failed to improve over the last 18 months. On Linux this isn't an issue as the kernel scheduler is clever enough to do this work. Hence the huge performance difference on the TR4 CPUs between Windows and Linux even on simple tasks like compressing/decompressing

 

[ToTheMax]

Dynamic mode looks excellent, removes the drawbacks and hassles of threadripper in Windows. In the words of Jensen "it just works".

 

[smogsy]

The paragraphs you have quoted are fine, but there is no performance gain whatever way you look and it and for me anyway the paragraph and chart damages the press release. Remove that and it's fine. They are claiming a performance gain when there simply isn't one irrespective of how they explain how it works. Even in respect to the rest of the article you could still get misled into thinking DLM is doing something extra from that paragraph.

whilst i see where your coming from, you thinking of it in the wrong way.

• your thinking stick it into game mode and you will get that performance anyway, however:
• most people wont be doing that, most people will spend 10% of their time in X app would benefit from game mode then 20% in X app that is better in creators mode and so on.
• Now No one in their right mind will reboot every time their application changes, (20 times a day for example) Thus when they do not reboot for X app they are losing performance which they could have gotten if they had switched for every app. this is where the dynamic mode comes in giving you a free performance boost on the fly by adjusting how the Processor/memory functions without rebooting.

@amigafan2003 makes sense right?

 

[Deleted member 66701]

Spot on smogsy.

Plus dynamic local mode is great for those users like me that game AND do something else concurrently, like 3d rendering.

 

[Panos]

@smogsy spot on mate


@LtMatt do you know if Dynamic Local Mode will work if a user has only dual channel ram on TR4 and not quad channel?
Asked the same question on the original blog post, but no reply. :/

 

[Vince]

@smogsy spot on mate


@LtMatt do you know if Dynamic Local Mode will work if a user has only dual channel ram on TR4 and not quad channel?
Asked the same question on the original blog post, but no reply. :/

Just to add to the questions will it work with 1st gen?